Page 1 of 1

self optimization

Posted: Wed Apr 07, 1999 4:12 pm
by "Ali Kerem Saysel"
In stock-flow language, do we have a generic structure describing self optimizing behavior. For example, think about a farmer monitoring only the two factors: pesticide application rate and yields. His strategy is to add more pesticides next year if his income is increasing by increased pesticides and to drop pesticide consumption if income is declining. Can we create a stock - flow formulation for this process.
Thank you all.
---------------------------------------
Ali Kerem Saysel, Ph.D.
Res. <b>ass</b>. in Environmental Sciences Inst.
Bogazici University, 80815 Bebek,
Istanbul, Turkey.
saysel@boun.edu.tr
tel: +90 212 2631540 - 2145
fax: +90 212 2575033

self optimization

Posted: Fri Apr 09, 1999 1:55 pm
by "geoff coyle"
Thats easy. Just create two levels for average income, smoothed over, say,
6 and 12 months (or whatever periods you choose, though be careful about
DT). Have an auxiliary for the difference between them and, depending on
whether the auxiliary is positive and negative, Bobs your uncle, as we say.
In practice, you might want to change the pesticide use if the auxiliary is
more than, say 10% of the smaller of the averages, so that you dont chase
noise.

Changing use next year is getting close to using a discrete formulation and
many people object to that. I dont because I want the model to do the same
things as the real system and for the same reasons, within the
simplification one has used.

Hope that helps. Give my regards to Yaman.

Regards,

Geoff Coyle
From: "Geoff Coyle" <Geoff.Coyle@btinternet.com>

self optimization

Posted: Wed Dec 08, 1999 9:41 am
by Nelson Repenning
I have a PhD student, John Foster, working on this problem. He is trying
to develop a fully behavioral model of how somebody might solve such an
optimization problem. It turns out to be quite a hard problem and we could
not find a formulation in the existing literature that really captured it.
Many of the economics models have a hill climbing formulation where the
decision maker assesses as the gradient of the production function and then
heads in the most desirable direction. However, all of these formulations
assume that the decision maker knows the functional form of the production
technology (perhaps with a significant delay). this does not seem like a
realistic assumption. we are working on a formulation in which the
decision maker calculates various covariations and uses that information in
a hill climbing formulation.

We be very interested in hearing from others who have tackled similar issues.

nelson




--------------------------------------------------------------
Nelson P. Repenning
Robert N. Noyce Career Development Assistant Professor
Operations Management/System Dynamics Group
Alfred P. Sloan School of Management, MIT
30 Wadsworth St, E53-339
Cambridge, MA 02139
phone: 617258-6889 fax:617258-7579 e-mail:nelsonr@mit.edu
http://web.mit.edu/nelsonr/www

self optimization

Posted: Thu Dec 09, 1999 9:42 am
by "John K. Blood"
Hello!

This may be way off base -- Im no mathematician, but some time ago I bought
a small book entitled, "Human Motivation and the Dynamic Calculus" authored
by Raymond B. Cattell, published by Praeger in 1985. Hardcover. ISBN:
0-03-072009-5

Perhaps you know of this book and the authors work? I had never seen
anything like it when I found it. The authors first article upon the
Dynamic Calculus appeared in 1935. This book "puts in a nutshell the whole
development of dynamic theory and the dynamic calculus," as written in the
authors bio. It contains many equations reflecting human behaviors and
attitudes, as well as explains how to put the equations to work. In 1975,
he co-authored with Dennis Child, "Motivation and Dynamic Structure." I
picked this text up on a whim with a feeling that someday my
non-mathematical self may find use for it connected with SD.

Your question piqued my curiosity and I remembered the lonely volume on my
human behavior shelf. Im wondering if this reference may be of assistance?
I am also wondering the value of what I have laying before me? Would any of
this be useful (even if for ideas) to the study of SD?

I hope this helps. Fascinating problem.

John Blood

Quest Dynamics, Inc.
jblood@quest-dynamics.com ** http://www.quest-dynamics.com
1-888-216-2916

self optimization

Posted: Fri Dec 10, 1999 11:49 am
by Tom Fiddaman
Christian Kampmans dissertation, "Feedback Complexity and Market
Adjustment: An Experimental Approach" (MIT, 1992), contains some examples
of hill-climbing behavior rules (setting price and quantity for profit
maximization). Its a bit thick, but I consider it a must-read for advanced
SD practice.

Tom

****************************************************
Thomas Fiddaman, Ph.D.
Ventana Systems http://www.vensim.com
8105 SE Nelson Road Tel (253) 851-0124
Olalla, WA 98359 Fax (253) 851-0125
Tom@Vensim.com http://home.earthlink.net/~tomfid
****************************************************